Ha yes, I’m just working this through for myself, or in case anyone else is as dense as me… 
I have to admire your eagle eye even catching this… I am however now tempted to start asking other questions about the engine that I was always a bit perplexed about like: Why does Harmonic+ affect both A and B1 - is there a reason this “works” in fm that makes it logical, or just a design decision?
1 Like
Actually that’s a lot of fun here!! 
3 Likes
It can be fun to turn knobs randomly or randomize parameters.
But if you want to recreate an FM patch with values, you need to understand how it works, and knowing what are operators values helps !
2 Likes
What could be more fun than understanding B operators!!!???!!!
3 Likes
Do you prefer B1 or B2 ?
B1! Why settle for 2nd best??
2 Likes
Oh…B1 of course…
I think there were a few decisions made during design phase in order to keep it simple and mitigate peak load on the cpu. These are all good questions! Many of them have been asked already and answered by @Ess and others on elektronauts when DN was launched and firmware got significant improvements. If you’re willing to search a bit, you’ll find true gems in other threads!
2 Likes
Thanks, I’ve been on the board since the DN launch so have probably read through it all years ago… but I sold it and just repurchased last year so have forgotten a bunch I’m sure.
I’ve always been interested in why certain operators are grouped in lfos, phase reset, harmonics, and if there is something there that might help me understand fm better, or if they were more decisions around cpu load as you say or other factors (maybe even… why not??).
Time to search…
2 Likes
Around value 16 
11 Likes
But… where? Value 85 doesn’t seem to be correct as lowest midpoint?
1 Like
So the answer to the Life the Universe and Everything is 16, but apart from that there are a few different tangles here I think it’s worth pedantically unravelling.
First, there’s how “level” is rather ambiguous with respect to the DN. In a lot of FM implementations, the “level” of an operator is both its “volume” and its modulation index. That is, if it’s a carrier, “level” affects how loud it is. If it’s modulating another op, “level” affects how much it modulates by.
In the DN, these concepts are decoupled. Level is exclusively linked to modulation index. This is why “level” doesn’t do anything when an op is a carrier — it’s only controlling modulation and a carrier has nothing to modulate.
With no “level” control over loudness, does that mean carriers always blast at full volume? No, there are a few things that can adjust this. First and foremost there’s the X/Y mix. By adjusting this you can make a carrier louder or softer in relation to another, and then use track- or amp-volume to make up the difference. But there are also envelopes to consider.
When looking at any of the algorithms on the DN, you might notice some lines are solid (usually ops modding other ops) and some of the lines are dotted (usually carriers going to X or Y). Solid lines have their envelops applied. Dotted lines do not.
You’ll note algos 7 & 8, though, have solid lines going to X and Y. This means the loudness of those op is controlled by their envelopes.
So to get to the idea of a four-bar organ patch, where you’d want four ops at ratios 1:1, 2:1, 3:1, 4:1, but at decreasing loudness, you could simply use algo 7. We don’t want anything modulating anything so set both A and B levels to zero. Then give C a ratio of 1, A a ratio of 2, and one B a ratio of 3 and the other a 4 (doesn’t really matter which). Now all are playing at their proper respective pitches!
But they’re also all equally as loud (assuming X/Y mix is centered). We want A to be less loud than C and the Bs to less loud than A. X/Y mix won’t help us because that would keep A and C equal in volume.
But note that A, B1, and B2 all have solid lines leading to X and Y. That means their envelopes will affect their volume. So we can set A’s env to a 0 atk and, say, 80 sus. Set B’s to 0 atk and 40-ish sus. There you go! Organ patch.
15 Likes
A will still modulate C in algo 7 despite being fed directly to the output. So there goes your theory.
Not if, as I said, you set the level to zero.
This is, indeed, the whole point. One might assume that setting level to zero would effectively mute A. It would on most other FM synths. But on the DN, level exclusively affects modulation index, not “loudness”. So by setting level to zero, we are only saying “don’t mod anything”. The “loudness” of A routed to X is controlled by its envelope in this algo (and we see that because the line from A to X is solid).
2 Likes
All true, very well explained – I’m pretty familiar with that.
Still I had something else in mind. If you want a pure additive patch with individual volume levels for all 4 sines and no modulation, you’ll not succeed in the DN but you could in the classic DX for example.
I am also very well aware that this is just a tiny niche. Not a huge missed opportunity. I like my DN the way it is.
Yes. To be clear, this gives you a purely additive patch with independent control over the frequency of the four sines and individual control over three of the four volumes. B1 and B2 share and env, so their volume is linked. But theirs can be different from A which can be different, again, from C.
Still, three outa four ain’t bad
But then, also, the DN will let you layer patches. So using two algo 7s you’d have a total of 6 ops with independent frequency and volume and could make any 6-op DX organ patch.
EDIT: well, not really. The DX has its unique envelopes to contend with. You could get close. And then maybe close the gap with LFOs? But yeah. The DN can do most FM-ish stuff, it just gets there differently.
5 Likes
Man…. You’re absolutely right! Never thought of the envelopes as volume controls independent from mod levels. So I actually learned one or two things from this thread! Awesome, thanks! 
3 Likes
Yeah the solid lines on algos 7&8 didn’t make much sense (well I got it but not so much WHY) until this thread untangled the level thing for me… going to try a few things this weekend with all of the above in mind.
2 Likes
Coming from non-Elektron FM (Yamaha, etc.), that graph locks my head.
What’s the thinking here? Why should raising modulation level for one of the B modulators reduce the modulation level for the other B modulator?
I’ve made grudging peace with not having 2 separate envelopes and 2 separate level controls for B1 and B2 (which is how any other civilized FM synth does it) … but what’s with this oppositional arrangement? Wouldn’t it make more sense for the B level control to raise or lower level (or modulation index, if you prefer) for both B1 and B2?
Not trying to be difficult, I truly don’t get it. Have programmed the DN anyway, but tending toward just using C, A, and one of the B ops if I want to understand what I’m doing.
2 Likes
Well if both just went up at the same time you would increase depth of modulation for both in parallel, this implementation actually, awkwardly, gives you more leeway: 1-43 is B1 only, then you crossfade between levels, finally at 85, B1 bottoms out at ~16 (we’ve learned) while B2 is always “full.” You probably already knew that but it does give more options then both just increasing both at once.
Looking at the algos B2 always mods B1 (when they both are mods*), so after 85 B1 is almost like a dam allowing how much of the full B2 goes through… it’s not perfect but it is more mallable, just… not intuitive.
I went DN->DX21->DN again - so I get where you’re coming from.
*5&6 do have B2 also modifying other ops
2 Likes